Chemical Engineering Science, Printed in Great Britain.

Vol. 42. No.

7. pp. 1585-1593.

1987. Q

0009~2509/87 1987 Pergamon

S3.00+0.00 Journals Ltd.

A DETAILED

MODEL FOR IN POROUS

NONISOTHERMAL ADSORBENTS

SORPTION

L. M. SUN and F. MEUNIER Laboratoire de Thermodynamique des Fluides, U-A. C.N.R.S. 874, Bat 502 ter Campus Universitaire, 91405 Orsay, France (Received 16 May 1986; accepted for publication 9 October 1986)

Abstract-A more detailed model for nonisothermal sorption in a single microporous particle is proposed, based on the gradient of chemical potential as driving force and simultaneous gas and adsorbed phase diffusion. The system of the simultaneous mass diffusion and heat conduction equations, linearized to simplify the analysis, is solved analytically, yielding the internal concentration. temperature profile and the kinetic uptake of the particle. It s shown that two parameters play an important role: Lewis number Le and thermal Biot number Bi. The isothermal model would be valid if LeBi > 100 and the uniform temperature profile model proposed by Lee and Ruthven would be a good one if L.e z- 10. In other cases, the more complete model presented herein should be used.

INTRODUCTION

The sorption in microporous adsorbents is always associated with large heat generation, and the influence may be all the more important since the diffusing process is rapid and the heat of sorption is high. However, the thermal effects were not taken into account until conflicting results for diffusivities obtained by the traditional transient uptake method and the Nuclear Magnetic Resonance (NMR) methods were observed. From then on, different models have been tested to analyze nonisothermal sorption in microporous adsorbents and it turned out that the intrusion of thermal effects may be one of possible explanations for discrepancies in measurements of diffusivities. Several models describing nonisothermal sorption in microporous adsorbents can be found in the literature. Chihara et al. presented a simplified model neglecting the internal heat and mass transfer resistances (1976) while the model of Lee and Ruthven assumed an infinite thermal conductivity so that the temperature was uniform throughout the adsorbent particle (1979). Based on this model, they investigated two different cases-intracrystalline diffusion control and bed diffusion control (1980, 1981). The same model was used by Kocirik for constant volumevariable pressure conditions (1984). Brunovska et al. and Ilavsky et al. proposed a simultaneous heat and mass transfer model for a porous spherical particle of finite thermal conductivity and the comparison between experimental and numerical results were presented (1978, 198Oa, 1980b, 198 I). A similar model was proposed by Haul and Stremming and an analytical solution was obtained. The models above cited made the following important assumptions: (i) A simple Fickian equation for mass diffusion is used, although it s known that the true driving force (when temperature gradients exist) is the gradient of chemical potential.

(ii) Only the gas phase diffusion is considered. In a more recent paper, we proposed (Sun et al., 1986) an analytical model using the pressure gradient as driving force and taking into account two phase diffusions--vapor and adsorbed phase diffusions. We propose herein a more detailed model for nonisothermal sorption in microporous adsorbents. This model will use the gradient of chemical potential as driving force and consider both vapor and adsorbed phase diffusion.
THEORY The model to be presented is based on the following

conditions: (i) The dominant mass transfer resistance is intracrystalline diffusion. (ii) The mass diffusion in the pore volume may occur by two different mechanisms: -surface diffusion of adsorbed molecules -pore diffusion of gas molecules. (iii) The chemical potential plays a similar role in governing the mass diffusion as temperature does for the heat diffusion so that the true driving force for mass diffusion is the gradient of chemical potential rather than that of concentration (Ruthven and Loughlin, 1971; Ruthven, 1984). The mass diffusion rate is related to the gradient of chemical potential through the mobility B: J, = - BmVp. (1)

This equation is applied to both the vapor phase and the adsorbed phase. (iv) Rates of adsorption are generally controlled by transport processes within the pore network rather than by the intrinsic kinetics of sorption at the surface so that we do suppose that the equation of state describing the equilibrium between the adsorbent and the vapor is valid. As an approximation, the vapor is considered as an ideal gas. The chemical potential of

1585

1586

L. M. SUN and F. MEUNIER where the total or effective diffusion coefficient D = D, is a function of concentration q and tempera+D, ture T.
Flux of heat dzffusion J,=

ideal gas can be defined by: (2)

q(T)

is defined here as a function of temperature:

(3)

-AVT

(11)

q(T)=h*-TS*+Cp

where ;1 is the effective thermal conductivity.

Mass balance for the two phases

where heat capacity of gas is considered as independent of the temperature. (v) The pore volume is divided up by the vapor phase c and the adsorbed phase q both expressed in kg per unit volume of the porous adsorbent. We may write: ++; = & (4) a where pv and pa are densities of vapor and adsorbed phases respectively, E is the porosity of porous adsorbent. The two phases having the same pressure:

Two mass balances are considered taking into account the diffusion terms as well as the sources in each phase:

ac z= - vJ,,+
&I t= --vJ,+

~2,

(12)

~oIa

I, and I, are mass source or sink resulting from phase change between vapor and sorbate so that the sum of these two terms is zero: Z,+I,= 0. (14)

pvRT=

P(q.T).

(5)

P(q, T) is nothing but adsorption isotherm which

can be defined by Dubinin s equation

P(q,T) = ..,,)exp~-~[~ln(~)Ilh)

The differential mass transfer equations will be of the form:

& =

(6)

with

Ps(T)

: the saturated pressure of liquid at temperature T

at

VCD,(Vq+GVT)]+p,l, +p,l,.

(151

a4 = V[D,(Vq+aVT)]

n, d : two coefficients.

at

(161

Thus, the vapor concentration c can be related to the adsorbed concentration q and the temperature T by the following equation:

The total mass balance is given by: g+$ = V[D(Vq+SVT)]. (17)

c=:,-P(q.T).

E-cl/P,

(7)

The mass source or sink terms I, and I, can be derived from eq. (15):
poIa= -p,I,=

(vi) The porous adsorbent is considered as a homogeneous medium for heat transfer and a single effective thermal conductivity is used for making use of the Fourier heat conduction equation. On the basis of these conditions, the kinetics of nonisothermal sorption in porous adsorbents can be described as follows.
Flux of mass diflusion J,,= -&cVp J 11111=-B,qVp

V[D,(Vq+NT)]-2.

(18)

Since c is a function of q and T, we may write: g It yields: (l+($)jz+(&),$=

V(D(Vq+SVT)} (20)

= (~)r$+(&)~~.

(19)

= -DJVq+bVT)
= -D,(Vq+SVT)

:;;

where

and
POI, = -p,l, = V[D,(V,+SVT)] -(g),$-(g),z. (21)

Heat

balance

the total flux is given by making summation:

J,=J,,+J,,= -D(Vq+SVT) (10)

The corresponding balance equation for heat transfer is obtained taking into account a diffusion term (corresponding to a homogeneous medium) and a

Nonisothermal

sorption in porous adsorbents

1587

source term corresponding to the phase transition: p&g = V(niVT)+p,l,AH (22)

where AH is the heat of adsorption. The introduction of the mass source term yields:

A @$+ (,,,Cp+AH(~)q)~= V(avT)

+ AHV [D,(Vq + SVT)]. (23) All transfer coefficients in this set of transfer eqs (20) and (23) are, a priori, functions of adsorbed phase concentration q and temperature T. Let us now present results obtained with that set of equations in a particular case. Numerical simulation We consider a spherical particle of porous adsorbent subjected to a small change in sorbate pressure at initial time. This case is of interest since it corresponds to an experimental procedure used to determine the mass diffusion coefficient. In order to simplify the theoretical analysis the following approximations will be used: (i) The stepwise changes of sorbate concentration and temperature are sufficiently small, thus all coefficients in transfer equations can be considered as independent of concentration and temperature. (ii) Equilibrium of sorbate at the particle surface is instantaneous. Based on these approximations, we obtain a simplified linear set of differential equations for heat and mass transfer:

where PI and Tr are pressure and temperature of the ambient fluid, eq. (28) represents pressure instantaneous equilibration condition and eq. (29) heat balance at the particle surface. These governing equations are made dimensionless by introducing the following variables and coefficients: Reduced variables:
q--40 w=----qm --4o

e=$

q=l ro

Dt r = -. r ,

Dimensionless coefficients:

su =

PI--PO

Cl = K(F1 G2= l+KF,

1)

qm is the final concentration of sorbate.

The eqs (24k(29)

are then written:

= a,+
+

~D,AHlp,C,
r2

a
(25)
r= 0

w=o
0 ae

e=l
zj qxo = 0

(32) (33) (34)

where a4 = I/p,C, is the thermal diffusivity. The appropriate initial and boundary conditions are as follows. Initial conditions:
t = 0: q = q,,

aw atf I=o =

WI,=, +cP(et,=l

-6,)

= su

and

T=

To.

(26)

Boundary conditions:

r=O
r = r,

ar

a4

0
r=#J =

CT
ar
r=O =

(27)

po+ (a4 >(41r=ro-40)+

T

ap

(gT>
4

(w,=ro--o)

= PI (28)

The important parameters in this set of dimensionless differential transfer equations are the thermal Biot number Bi = hr/d, representing the ratio of thermal internal resistance and external surface resistance; /3 = DJDa ratio of vapor phase diffusion coefficient and total diffusion coefficient and the Lewis number I_.e= as/D-rate of heat conduction/rate of mass diffusion.

1588 Examples of simulation

L. M.

SUN

and

F. MEUNIER

The set of linear partial differential eqs (30)-(35) is solved analytically by the Laplace transformation method. The analytical solution is given in the Appendix. In the simulations reported here, a reference case corresponding to the sorption of water into a single spherical particle of zeolite is chosen: r0 = 1 mm, E = 0.32 and pOC, = 2.0 x 106. Estimated values for the effective conductivity and the overall heat transfer coefficient: ;1= 0.3 W/m K and h = 30 W/m K. The initial and final conditions: To = 80C PO = 30mb and T, = 8OC, PI = 50mb. The two coefficients in the Dubinin s equation: n = 2 and d = 1.8023 x 10e7. The heat of sorption AH can be determined from adsorption isotherms (AH = 2.78 x 106). As to the diffusion coefficients D,, D, and D,the following values are (D, = D, = 10-5cm2/s). cm2/s used: D = 2 x lo- Finally, the entropy at standard state of ideal water vapor is S, = 10465. The above evaluated parameters yield: a = - 15, K = 8.7 x lo- , Bi = 0.1, Le = 75, y #I = 0.5, = 5.7 x 10e4, 4 = 25.6. It has to be noted that the value of the Lewis number is quite large due to the low value of the diffusivity.

A parametric study is now presented in which the values of some parameters are varied so as to determine their influence on the results of the numerical simulation.
RESULTS AND DISCUSSIONS

In Fig. 1, we show concentration and temperature profiles within the particle for the reference case. We may note that gradients of concentration are relatively important with regard to temperature gradients. In fact, the high value of Lewis number (IX = 75) indicates a rapid heat transfer and a relatively slow mass diffusion process. We may observe that, after an increase in a short time, temperatures decrease (almost linearly) due to the heat transfer with the ambient fluid. The maximum dimensionless temperature rise in this case is approximately 0.016 (corresponding to 56C).
Efict

of Bifor

different

Le

Figures 2 and 3 present the effect of thermal Biot number on volumic averaged concentration and surface temperature for Le = 75 and 7.5 respectively.

1 .oo rx)/..,/.,. 0.90 0.80 ,.., _,.

,_ __-. ;y~~;;T*Y ___---,A

-.w =~=

~-Y-.Aw.*

0.70~~~

,i ,/

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9 n

1.0

0.0

0.1

0.2

0.3

0.4

0.5 Ia)
Bi = 0

0.6

0.7

0.8

0.9

1.0

e
1.016

l.OlO_l

/-

1.018-

--__ ,, .\ \ .,

1.04/
1.006 1.004

2s
0.2 0.3 0.4 0.5 b) 0.6 0.7 0.8 0.9 9 1.0

--._

0.01 --._ ---___ --__

1.0161.0?41.012-

l.oool-_~r
0.0 0.1

Fig. 1. doncentration (a) and temperature (b) profiles. (Xx = 75, fl = 05, Bi = 0.1) for 5: (1) 0.0001, (2) O.ooO5, (3) 0.001, (4) 0.0025, (5) 0.005, (6) 0.0075, (7) 0.01, (8) 0.033, (9) 0.066, (10) 0.1.

(b) Fig. 2. Effect of Biot number on (a) averaged concentration and (b) surface temperature (Le = 75, 13= 0.5).

Nonisothermalsorption in porous adsorbents

w

1589

0.30 -I I 0.20 0.10 i

(a)
Bi = 0

O.OO VT_
(a)
1.0824-I

2 b)

2 lb)

Fig. 3. Effect of Biot number on (a) averaged concentration and (b) surface temperature (Le = 7.5, @ = 0.5).

Fig. 4. Influenceof r9 (ratio of vapor phase diffusivityand total diffusivity) on (a)the uptakeand (b) surfacetemperature I. (IX = 7.5, Bi = 0.1). -~=o,---~=o.5,---~B

From these figures, we can see: (i) For a small heat transfer coefficient h, the process may be controlled by thermal diffusion. In this case, an isothermal (Bi = co) model would yield erroneous results. (ii) In order that the isothermal condition be justified for a given diffusion process, the value of h must be sufficiently high. A criterion is proposed here: sorption may be assumed isothermal if LeBi > 100 (for example Bi > 1.4 for Le = 75).
Effect of j3 j?, the ratio

Fig. 5 for Le = 7.5. Notable differences due to the0 conductivity may be observed only for very small Lewis numbers.
Eflect of the driving force for massfiow The effect of the choice of the driving force for mass flow is shown in Fig. 6. The curves presented are

of gas phase diffusivity and total diffusivity, has not an important influence on kinetics of sorption for moderate Le, as we can see from Fig. 4 (the total diffusivity D is the same for different values of p). This influence becomes negligible if Le Z= 10, this means that we can take a single effective diffusivity if Le> 10.
Eflect of conductioity

obtained with the present model and with another model in which the driving force for the mass flow is the pressure gradient instead of the chemical potential gradient. The larger the difference, the smaller the Lewis number. For large Lewis numbers (Lc > 10) the difference becomes negligible.
Effect of the size of the particle

The influence of conductivity increases when the Lewis number decreases, an example is presented in

When the size of the particle is modified, the heat transfer between the particle and its surroundings is modified. The curves presented on the Fig. 7 have been obtained taking the same value (per square meter) for the heat exchange coefficient. It is seen that the smaller the particle, the larger the effects due to temperature inhomogeneity; for a particle size larger than 30mm

L. M. SUN and F.

MEUNIER

0
tl _ 1 .ozti 1.024 -

o.oo* , , ,
1 2 (a) 3 4 I 5 e 1.024 0

,, 1

, ,

, , , ,, 2

,, ,,

,, , 3

, ,,

, 4

, I

, 5

(a)

2 (bl

1.0007, 0

,, ,

,, , ,, 1

, ,,, ,

,, 2

,,

, ,, ,, , 3

,,

, , ,, 4

, *

, 5

(b)

Fig. 5. Effect of the effective conductivity. (a) averaged concentration; (b) surface temperature (Le = 7.5, fi = 0.5) L = 0.03, --a = 0.3, --A = 3.

= 0.1). (a) averaged concentration; (b) surface temperature. gradient of chemical potential as driving force. --gradient of pressure as driving force.

Fig. 6. Comparison of different driving forces (Le = 7.5, Bi

diameter in the reference case, the results are nearly the same as those obtained with an isothermal model. So, criteria for different models, as illustrated in Fig. 8, can be defined as follows: IX > l&--uniform temperature profile model LeBi > lO&-isothermal model other-present model (or similar model).
CONCLUSIONS

The model presented here provides a more detailed analysis for non isothermal sorption in a single microporous particle taking into account the influence of temperature gradients within the particle. The important parameters of this model are: Lewis number (ratio between thermal diffusion rate and mass diffusion rate), and thermal Biot number (ratio between internal and external heat transfer resistances). Two conclusions can be inferred: 0 when Le > 10, the heat transfer inside the particle is so rapid that the uniform temperature profile may be assumed. 0 when the size of the particle is increased, the Biot number increases at constant Lewis number, so that we go into the isothermal domain (see Fig. 8).

The results obtained with the present model are compared to those obtained with simplified models. It is shown that the kinetics of sorption can be described by one out of three models depending on the values of the Lewis and Biot numbers (Fig. 8). The limits between the domains where the different models are valid are defined as follows: l the isothermal model is valid if LeBi > 100, in that case, only one parameter D is necessary to describe the kinetics of the sorption. m the uniform temperature profile model is valid if L.e > 10; two parameters D and h are necessary to describe the kinetics. m out of those limits, the present model should be used; in that case, the gradient of the chemical potential has to be used as the driving force. Two diffusion coefficients corresponding to gas and adsorbed phase diffusion are necessary. Thus, four parameters D,, D,, h and A are necessary. To observe directly experimentally the effects of a non uniform temperature profile, experiments have to be performed at very short times after either a pressure step or a temperature step in which the temperature must be measured in several points within the particle.

Nonisothermal

sorption in porous adsorhents

1591

0.00

, , / , , , ,
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

T
0.9 T &

Fig.

8. Illustration

of

limiting

mode&.

conditions

for

different

D Da D h h* 1, 1

0.0

0.1 0.2

0.3

0.4 0.5
(b)

0.6

0.7

0.8

0.9

1.0

Fig. 7. Influence of the particle size on the kinetics (Le = 75, jl = 0.5). (a) averaged concentration; (b) surface temperature r0 = lmm, --r0 = 2.5, --r0 = 5. ---Bi = cc (isothermal case).

J,

The pair adsorbent-adsorbate has to be selected so that it corresponds to a Lewis number less than 10: the effect will be all the more important since the diffusing process is rapid, the size of the particle is small and the heat of sorption is high. This model can be applied only to monodisperse adsorbents but it is being extended in our laboratory to bidisperse adsorbents.
supported by Acknowledgements-This research was C.N.R.S.-P.I.R.S.E.M. (Programme lnterdisciphnaire de Recherches sur les Sciences pour 1 Energie et les Mat&es Premieres) and A.F.M.E. (Agence Franqaise pour la Maitrise de 1 Energie). NOTATION a4 B thermal diffusivity, m2/s

J ma J mv

J, K Le Le 111
n P P* PO Pl P, 4 40 4 r R S su ; TO T,

4 Bi 4

mobility, s mobility of the adsorbed phase, s Biot number mobility of vapor phase, s vapor phase concentration, kg/m effective heat capacity, J/kgK constant in the Dubinin s equation

total diffusivity, m2/s adsorbed phase diffusivity, m2/s vapor phase diffusivity, m2/s overall heat transfer coefficient between adsorbent particle and ambient fluid, w/m K enthalpy of vapor at reference state, J/kg mass source of adsorbed phase, s-l mass source of vapor phase, s-l total mass flow density, kg/m2 s mass flow density of adsorbed phase, kg/m2 s mass flow density of vapor phase, kg/m2 s heat flow rate, w/m2 dimensionless parameter defined in text Lewis number dimensionless parameter defined in text concentration, kg/m3 constant in Dubinin s equation sorbate pressure, Pa pressure of vapor at reference state, Pa initial pressure, Pa pressure of ambient fluid, Pa saturated pressure, Pa adsorbed phase concentration, kg/m3 initial adsorbed phase concentration, kg/m3 final adsorbed phase concentration, kg/m3 radial coordinate, m adsorbent particle radius, m gas constant, J/kg K entropy of vapor at reference state, J/kgK dimensionless parameter defined in text time, s absolute temperature, K initial temperature, K temperature of ambient fluid, K

1592

L. M. SUN and F. MEUNIER soumis P un 6chelon de pression de vapeur adsorbable. lnt. J. Heat Mass Transfer 23, 1393.

Greek symbols dimensionless parameter ratio of vapor phase diffusivity and total ; diffusivity dimensionless parameter Y AH heat of adsorption, J/kg porosity of adsorbent particle effective thermal conductivity, W/m K 1 chemical potential J/kg P normalized radial coordinate (r/r ,), function rl of temperature defined by eq. (3) CO normalized adsorbed phase concentration ((4 -qo)/(qm --4o)) w normalized adsorbed phase concentration averaged over a particle density of adsorbed phase, kg/m3 Pa density of vapor phase, kg/m3 P density of solid, kg/m3 PO normalized time variable (Dt/ri) z normalized temperature variable (T/ To) 8 Subscripts a adsorbed phase V vapor phase 0 initial state co final state
REFERENCES

APPENDIX The linear differential system formulated in the text (eqs 30-35) can be solved by use of Laplace transformation method. After simple algebraic manipulations, we may obtain from eqs (30) and (31)

where aI 1 = (BFz K - G&Q

012 = (Fz JX -a&)/Q a22 = (G --FL m)/Q

~21 = (GI -BF,

Q=F2G,-FIG*.

K)IQ

The I-aplace form of these equations is: (A.3) (A.4) The corresponding equations for boundary conditions are:

(A.3

Brunovska,A., Hlavacek, V., Ilavsky, J. and Valtyni, J., (a) 1978. An analysis of a nonisothermal one-component

sorption in a single adsorbent particle. Chem. Engng Sci. 33, 1385. (b) 1980, Non isothermal one component sorption in a single adsorbent particle, effect of external heat transfer. Chem. Engng Sci. 35,757. Brunovska, A., Ilavsky, J. and Hlavacek, V., 1981, An analysis of a nonisothermal one component sorption in a single adsorbent particle-a simplified model. Chem. Engng Sci. 36, 123.. Chihara, K., Suzuki, M. and Kawazoe, K., 1976, Effect of heat generation on measurement of adsorption rate by gravimetric method. Chem. Engng Sci. 31, 505. Haul, R. and Stremming, H., 1984, Nonisothermal sorption kinetics in porous adsorbents. .J. Colloid Interface Sci. 97, 348. Ilavsky, J., Brunovska, A. and Hlavacek, V., 1980, Experimental observation of temperature gradients occurring in a single zeolite pellet. Chem. Engng Sci. 35, 2475. Kocirik, M., Struve, P. and Bulow, M., 1984, Analytical solution of simultaneous mass and heat transfer in zeolite crystals under constant-volume/variable pressure conditions. J. Chem. Sac. Faraday Trans. I 80, 2167. k, L. K. and Ruthven, D. M., 1979, Analysis of thermal effects in adsorption rate measurements. J. Chem. Sot. Faraday Trans. I 75, 2406. Ruthven, D. M. and Loughlin, K. F., 1971, Correlation and interpretation of zeolitic diffusion coefficients. Trans. Faraday Sot. 67, 1661. Ruthven, D. M., Lee, L. K. and Yucel, H., 1980, Kinetics of nonisothermal sorption in molecular sieve crystals. A.1.Ch.E. J. 26, 16. Ruthven, D. M. and-Lee, L. K., 1981, Kinetics of nonisothermal sorption: systems with bed diffusion control. A.I.Ch.E. J. 27, 654. Ruthven, D. M., 1984, Principles of Adsorption and Adsorption Processes. Wiley-Interscience. Sun, L. M., Meunier, F. and Mjschler, B., 1986, Etude analytique des distributions de temp&rature et de concentration g l intkrieur d un grain sphkrique d adsorbant solide

The solution of this set of differential equations in Laplace form is described as follows:

8=
where G.2 =

--&CP.WsWhsJ;)_P~(s)sh(hv&)l
al~+a22~C(a11+~n22)2-44(a11a22-a12a21)7Z
2hla22-a~2a2l)

(A.91

EI = -l+(A,+$)y+Bi F, = bl(Bi-Eel)
el = EL - Bi ;u=I =

Ez=

-l+(A,++)u+Bi

FZ = b2(Bi-EEl) ) (A, + 4) e2 = E2Bi s: ) (A2 + 4)

Fz(A1+4)

=z = FICA,+&)
G = &(AI+~)-EI(Az+~)

PI@) = eLsh(bl&)+F,&ch(bl&) Pz(s) = ezsh(b&)+F,&ch(b,&)

Q(S) = Gsh&&)sh(b&) +z&sh@&)ch(b,fi) --&ch(b,&)sh(b&).

Nonisothermal

sorption in porous adsorbents

1593

The expressions for sorbate concentration and temperature evolution in the real time domain are obtained by use of residue inversion method: o=l+E -&PI E=e,+ m 2Suexp(--pit) n=, rlp,Q (P,) {AI PZ (p,)sin (bt ftpn) (A.10) tPz (p,)sin (b, VP) (A.ll)

Qb.1

= C% -zdcos
+P&I~, - ~,(z~b~

hp,)sin

&PO)

+ (Gbt + =I ) sin (b, P,) cos (b, P,,) -z,b,)cos(b~~,)cos(bzp,) -zzbl)sin (b,p,)sin (b,p,).

(p,)sin(bzqp,,)l m 2Suexp(-pir) c VP~Q (P,) =I

P, is given by the roots of: Gsin (blp,)sin (b,p,) + zlposin - z,p,cos (bl~,)cos (b,p,)sin (bzp,,) (b2pn) = 0. (A.12)

- PI (P,) sin (bzqp,)> where

The sorbate concentration averaged over the particle is

,=1+x m =I

pi=

--s
+ (bt~,)

sin (b, P,) (hip,)* (P,) > (A.13)

P, (p,) pz (p,)

= cl sin (blp,)+F1~,cos

- &PI

= ez sin (bzpn) + FZP~COS (bzp,)